Plant interactions play an important role in shaping the composition and structure of plant populations and communities (Baer et al., 2003; Rajaniemi et al., 2009; Wassmuth et al., 2009; Cahill and McNickle, 2011; Houseman, 2014; Zhang et al., 2020). Numerous studies have explored plant interactions (Robinson et al., 1999; O’Brien et al., 2007; Allesina and Levine, 2011; Roiloa et al., 2014; Ravenek et al., 2016; Fichtner et al., 2018; Garlick et al., 2021), yet, several factors impede researchers to exactly quantify such interactions. For example, it is difficult to connect the cause and effect of plant interactions in natural communities with multiple species growing together, where many other factors such as climate change, disturbance, and natural heterogeneity may affect the results (Ravenek et al., 2016; Liu et al., 2019). Previous controlled studies often explored plant interactions in pairs for simplification (Johansson and Keddy, 1991; Chesson, 2000; Day et al., 2003; Rajaniemi, 2011; Li et al., 2018). However, plant individuals interact not only in pairs but are also affected by higher-order patterns (i.e., the interactions between two species are likely to be modified by other species, Pierik et al., 2013; Grilli et al., 2017; Levine et al., 2017; Fichtner et al., 2018; Li et al., 2020, 2021; Chu et al., 2021). Thus, to quantify plant interactions more accurately, it is necessary to quantify the differences between higher-order interactions and pairwise interactions.

Natural soils are heterogeneous, and this soil heterogeneity has two components, namely, qualitative heterogeneity and configurational heterogeneity (Kelly and Canham, 1992; Maestre and Cortina, 2002; Wijesinghe et al., 2005; Maestre and Reynolds, 2007). The former refers to the variations of texture, nutrients, pH, etc., between patches in the soils, while the latter reflects the size of these patches (Dufour et al., 2006). Soil heterogeneity in this study refers to the qualitative heterogeneity. Qualitative heterogeneity affects plant interactions by altering the availability of soil resources such as water and nutrients (Fransen et al., 2001; Schenk, 2006; Zhang et al., 2020).

Plants growing in a population or a community interact with each other (Chesson, 2000; Levine et al., 2017; Garlick et al., 2021). However, how to quantify such interactions is a hot topic in ecology. Many parameters can be applied to quantify plants interactions (Weigelt and Jolliffe, 2003), where relative competition intensity (RCI) is a widely used one. Previous studies found that plant interactions vary with species identity (Catorci et al., 2011; Raath-Krüger et al., 2019) and their growing stages (Yang and Rudolf, 2010; Li et al., 2019). Plant interaction at the early growing stage is crucial for plant establishment in plant populations and communities (Lortie and Turkington, 2008; Hart et al., 2018), and warrants more research. Plant height is a good indicator of plant interactions since it plays a vital role in determining light interception (Xiao et al., 2007), and thus, it is widely applied to quantify plant interactions non-destructively (Cui et al., 2004; Chen and Nelson, 2006; Xiao et al., 2007).

To explore the effects of soil heterogeneity and species identity on plant interactions, a controlled experiment was conducted, where three levels of soil heterogeneity were developed by mixing local soil and sand in three different ratios (i.e., soil:sand ratio = 2:8, 5:5, and 8:2), and three typical forage grasses were used (Festuca elata, Bromus inermis, and Elymus breviaristatus). RCI was applied to quantify plant interactions, where RCI was calculated as a direct ratio of monoculture and mixture performance, and positive (or negative) values simply indicated that a plant individual experiences stronger (or weaker) competition in mixtures compared with growing alone (Weigelt and Jolliffe, 2003). Specifically, we put forward three hypotheses: (i) RCI is expected to increase with increasing soil heterogeneity, where higher soil heterogeneity could reduce plant interactions due to an increase in the available resources at higher soil heterogeneity levels (Levine et al., 2017; Liu et al., 2017b); (ii) Such patterns may be modified by plant species identity since responses to soil heterogeneity can be modified via strategies or traits such as plasticity (Hodge, 2004; Craine and Dybzinski, 2013); (iii) Soil heterogeneity and species are assumed to jointly affect RCI since plant responses to soil resources depend on their neighbor identity and the distribution pattern of soil resources (Mommer et al., 2012; Garlick et al., 2021).

Soil–sand ratio and species × soil–sand ratio significantly affected RCI₁ (Table 2 and Figure 3). For mixtures with two individuals, RCI₁ of F. elata and B. inermis was highest at the medium soil–sand ratio, while RCI₁ of E. breviaristatus was highest at high soil–sand ratio (Figure 3A). For mixtures with three individuals, similar patterns were found (Figure 3B). However, species as a single factor did not affect RCI₁ (Table 2).

Species, soil–sand ratio × species significantly affected RCI₂ (Table 3 and Figure 4A). Specifically, RCI₂ of F. elata was higher in medium soil–sand ratio, while RCI₂ of B. inermis and E. breviaristatus increased with increasing soil–sand ratios (Figure 4A). However, soil–sand ratio as a single factor did not affect RCI₂ (Table 3). The separated analyses demonstrated that soil–sand ratios significantly affected RCI₂ of F. elata (Table 4). Specifically, RCI₂ of F. elata was higher at the medium soil–sand ratio, while RCI₂ of B. inermis and E. breviaristatus increased with increasing soil–sand ratios (Figure 4B).

Furthermore, species, soil–sand ratio × species significantly affected RCI_diff (Table 3 and Figure 4B). However, soil–sand ratio as a single factor did not affect the RCI_diff (Table 3).

We found that plant interactions were affected by soil–sand ratio, which differed among plant species. Specifically, RCI (RCI₁, RCI₂, or RCI_diff) of F. elata and B. inermis was the highest in the medium soil–sand ratio, while RCI₁ of E. breviaristatus was the highest in the high soil–sand ratio.

Our first hypothesis, stating that RCI would increase with increasing soil heterogeneity (i.e., soil–sand ratio), was supported. We found that soil–sand ratio significantly affected RCI₁ (Figure 3 and Table 2), which is likely caused by the impacts of soil heterogeneity on the root foraging and distribution of plants (Levine et al., 2017; Liu et al., 2017b). However, root performance was not measured at this stage as harvesting roots affects plant growth at the following stages. This finding (i.e., RCI increased with increasing soil heterogeneity) is consistent with previous studies, such as that of Yu et al. (2019) who found that soil heterogeneity increased plant competition intensity within a population at a relatively low plant density. Such positive effects of soil–sand ratio on RCI could be caused by the indirect effect of soil–sand ratio on mycorrhizal fungi, where the effects of mycorrhizal fungi on plant performance increased with increasing soil–sand content (Zaller et al., 2011). Yet, other studies reported that plant competition was affected by resource heterogeneity only when plant individuals were not genetically identical (Day et al., 2003; Zhou et al., 2012). Furthermore, the positive effect may disappear at the later growing stage of plants when the soil–sand ratio was modified by these plants during the growing stages (Roiloa and Retuerto, 2006; Dong et al., 2015). Moreover, such an effect may disappear in a population or community since some individuals in this population or community could detect the neighbors and avoid direct competition (Novoplansky, 2009; Lepik et al., 2021), and some other individuals may perform in the opposite direction. The result depends on the combination of these two effects. Thus, we assumed that both positive and negative effects on plant competition are present simultaneously along the gradient of soil–sand ratio. Generally, plants adopt a diversity of responses to environmental variation (Lawson et al., 2015). However, the potential mechanisms of soil heterogeneity on plant interactions merit further investigation.

Our second hypothesis assumed that the effects of soil–sand ratio on RCI could be modified by plant species identity. In line with our expectation, species significantly affected RCI₂ and RCI_diff (Table 3). This is consistent with previous studies that grasses-mixture had higher shoot biomass than root biomass (Bessler et al., 2009). Specifically, we found that in mixtures with two individuals, the medium soil–sand ratio supported higher RCI₂ of F. elata than the other two ratios. This may be related to differences in the electrical conductivity (EC), where the medium soil–sand ratio had a higher EC value than the other two ratios (Table 1). However, the high soil–sand ratio supported a higher RCI₂ of B. inermis and E. breviaristatus than the other two ratios (Figure 3A). This may be related to pH, where the high ratio had the highest pH value (Table 1). Yet, the underlying mechanisms of these two different patterns merit further study.

Plant responses depend on neighbor identity and resource distribution (Mommer et al., 2012; Garlick et al., 2021). Yet, it refers to the same species but different individuals in this case. Thus, we expected that soil heterogeneity and species would jointly affect plant interactions. This was supported since significant interactions between soil–sand ratio and species were found (Table 2), in line with previous studies. For instance, Nakamura et al. (2008) found that soil heterogeneity affected larger individuals but not smaller individuals, where smaller individuals were strongly impacted by their neighbors (Chen et al., 2020). Moreover, the effects of soil heterogeneity on plant height could also be caused by the different growth rates of these individual plants (Hutchings, 1997), where plants growing in patches with more resources tend to grow faster. Such results could be derived from the different germinations at the beginning of the experiment, where Liu and Hou (2021) found that the heterogeneous distribution of soil resources affected seed germinations, which might further affect their heights. Plant heights of the target species in this study may differ after germinations even plants with the similar size were transplanted in this study since the earlier germination seeds may have a high chance to access light, which would be much higher in the later growing stage. However, the lack of germination information of these species in this study impedes us to further test such a hypothesis.

Results of this study should be interpreted and extrapolated with caution due to the following reasons. (1) This was a short-term experiment with only three soil–sand ratios, which impedes us to make comprehensive conclusions. Thus, further research should include a series of soil–sand ratios in longer term experiments. (2) Previous studies found that natural conditions include diverse plant functional groups. However, only three grass varieties were applied in this study. Therefore, future studies should consider competitive exclusion or a wider range of species, including clonal and N-fixing species (Mayfield and Levine, 2010), where the former can improve their nutrient-use efficiency via clone integration (Ying et al., 2018), and the latter can modify soil conditions through increasing soil nitrogen by fixing nitrogen from the air (Carlsson and Huss-Danell, 2003; Bhandari et al., 2020). (3) Plant interactions are affected by many factors, which is why drivers (e.g., climate) should be considered (McKane et al., 2002; Harpole and Tilman, 2007). (4) Similarly, other physiological and morphological traits such as specific leaf area should be considered in calculating plant interactions (Janecek et al., 2004). (5) Plant interactions vary in time, so results in this study at the early growing stage may not be applicable for the following growth stages (Li et al., 2019).

This is, to our best knowledge, likely to be the first time to quantify the indirect interaction of plants in a controlled experiment, and such interaction is modified by the joint effect of soil heterogeneity and species identity. Natural soils include both qualitative heterogeneity and configurational heterogeneity. However, this study only considered qualitative heterogeneity. Future studies should take the configurational heterogeneity, especially in three dimensions, into account (Liu et al., 2017a,b, 2021). Moreover, only aboveground competition intensity was explored here, future studies should consider plant interactions belowground as plants having a size-based competitive advantage aboveground may not have the same competitive advantage belowground and vice versa (Casper and Jackson, 1997; Poorter and Nagel, 2000; Brown et al., 2019; Liu et al., 2021).

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

YL designed the study and wrote the first draft of the manuscript. GL, MW, and YL conducted the experiment, collected the data, and conducted the analyses. All authors contributed substantially to this study.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.